Large-scale air mass characteristics observed over the remote tropical Pacific Ocean during March-April 1999: Results from PEM-Tropics B field experiment

. Eighteen long-range flights over the Pacific Ocean between 38øS to 20øN and 166øE to 90øW were made by the NASA DC-8 aircraft during the NASA Pacific Exploratory Mission (PEM) Tropics B conducted from March 6 to April 18, 1999. Two lidar systems were flown on the DC-8 to remotely measure vertical profiles of ozone (03), water vapor (H20), aerosols, and clouds from near the surface to the upper troposphere along their flight track. In situ measurements of a wide range of gases and aerosols were made on the DC-8 for comprehensive characterization of the air and for correlation with the lidar remote measurements. The transition from northeasterly flow of Northern Hemispheric (NH) air on the northern side of the Intertropical Convergence Zone (ITCZ) to generally easterly flow of Southern Hemispheric (SH) air south of the ITCZ was accompanied by a significant decrease in 03, carbon monoxide, hydrocarbons, and aerosols and an increase in H20. Trajectory analyses indicate that air north of the ITCZ came from Asia and/or the United States, while the air south of the ITCZ had a long residence time over the Pacific, perhaps originating over South America several weeks earlier. Air south of the South Pacific Convergence Zone (SPCZ) came rapidly from the west originating over Australia or Africa. This air had enhanced 0 3 and aerosols and an associated decrease in H20. Average latitudinal and longitudinal distributions of 0 3 and H20 were constructed from the remote and in situ 03 and H20 data, and these distributions are compared with results from PEM-Tropics A conducted in August-October 1996. During PEM-Tropics B, low 03 air was found in the SH across the entire Pacific Basin at low latitudes. This was in strong contrast to the photochemically enhanced 03 levels found across the central and eastern Pacific low latitudes during PEM-Tropics A. Nine air mass types were identified for PEM-Tropics B based on their 03, aerosols, clouds, and potential vorticity characteristics. The data from each flight were binned by altitude according to air mass type, and these results showed the relative observational frequency of the different air masses as a function of altitude in seven regions over the Pacific. The average chemical composition of the major air mass types was determined from in situ measurements in the NH and SH, and these results provided insight into the origin, lifetime, and chemistry of the air in these regions. for each identified and these were with altitude in the hydrocarbons, C2H2/CO


Introduction
The NASA Pacific Exploratory Mission (PEM) Tropics B (PTB) was conducted from March 6 to April 18, 1999 (DC-8 portion of PTB) to investigate the atmospheric chemistry of with a vertical resolution of 300 m and an averaging time of 5 min (about 70-km horizontal resolution at DC-8 ground speeds) [Browell, 1983;Browell et al., 1983Browell et al., , 1985. Intercomparisons between in situ and UV DIAL 0 3 measurements are made throughout the course of a field experiment to insure the accuracy of the measurements is being maintained [see, e.g., Fennet al., 1999].
The LASE system was initially designed for and operated from the NASA high-altitude ER-2 aircraft [Browell and Ismail, 1995;Browell et al., 1997;Moore et al., 1997]. This system uses a Ti:sapphire laser that is pumped by a double-pulsed, frequency-doubled Nd:YAG to produce laser pulses in the 815-nm absorption band of H20. The wavelength of the Ti: sapphire laser is controlled by injection seeding with a diode laser that is frequency-locked to a H20 line using an absorption cell. Each pulse pair consists of an on-line and off-line wavelength for the H20 DIAL measurements. The LASE system was operated for the first time from the NASA DC-8 during the 1998 Convection and Moisture Experiment (CAMEX 3), which was conducted to study hurricane characteristics with a large ensemble of airborne instruments [Fen'are et al., 1999;Browell et al., 2000]. During CAMEX 3, LASE was configured to simultaneously measure H20 , aerosols, and clouds below and above the DC-8 using a back-to-back transmitter-receiver arrangement similar to that used for the UV DIAL system. This enabled LASE measurements to be made from near the surface into the upper troposphere.
The accuracy of LASE H20 profile measurements was shown to be better than 6% or 0.01 g kg -•, whichever is greater, in a major validation experiment conducted during 1995 . During this experiment, two validation aircraft (C-130 and Lear jet) were used to provide in situ H20 measurements along the LASE flight track, and one also had a H20 DIAL system for remote measurements along the same path. In addition, special radiosonde balloons were launched, and a ground-based Raman lidar was operated for H20 comparisons with LASE. Intercomparisons were also conducted in 1996 during the Tropospheric Aerosol Radiative Forcing Observational Experiment (TARFOX) field experiment [Fen'are e! at., 2000a, 2000b]. Once the LASE system was configured for operation from the DC-8 in the simultaneous nadir and zenith modes of operation, additional comparisons were made with other sensors to insure the accuracy of the H20 measurements. Fen'are et al. [1999] and Browell et al.
[2000] discuss these intercomparisons in the context of CAMEX 3. The results were consistent with the initial LASE validation experiments. An example of comparisons between LASE nadir and zenith H20 profile measurements and in situ H20 measurements from three different H20 sensors on the DC-8 and a locally launched radiosonde is shown in Plate 1. Although the sampling between the various sensors is very different, there is good agreement between measurements. The greatest disagreement occurs in the two-stage chilled mirror hygrometer measurements at H20 levels below 2 g kg -• during ascent where the instrument is constrained by its maximum thermoelectric cooling capacity. The radiosonde launched from Tahiti generally agrees with the other measurements with the largest deviation in the midtroposphere of -0.15 g kg -•. After the cryogenic and diode laser hygrometer measurements were smoothed to the same 500 rn vertical resolution of the LASE measurements, they showed excellent agreement with each other and with the LASE measurements. This further demon-strates the measurement capability of LASE across a wide range of H20 concentrations.
Since the UV DIAL remote 03 profiles begin about 750 m above and below the aircraft, the ozone distribution across this region is estimated by interpolating between the nadir and zenith UV DIAL measurements with the in situ 03 measurement used to constrain the interpolation in between. Ozone distributions were also extrapolated from the lowest UV DIAL measurement altitude (usually below 1 km) to the surface using the gradient in ozone concentrations determined from the in situ measurements. These techniques were previously discussed for the UV DIAL 0 3 distributions from the PTA field experiment [Fennet al., 1999]. An example of the resulting 03 distribution is shown in Plate 2 for the flight from NASA Dryden to Hawaii (flight 5). The aerosol scattering ratio distribution derived from the UV DIAL measurements on the same flight is also shown in Plate 2. The aerosol scattering ratio is the ratio of the amount of backscattered laser light from aerosols alone to the amount of backscattered laser light that would be expected from a clean (aerosol free) molecular atmosphere. The relative IR (1064 nm) lidar backscatter distribution, which represents scattering from aerosols and molecules, is normalized to a region that is estimated to be relatively clean (less than 2% of scattering attributed to contribution from aerosols), thereby yielding the total atmospheric scattering ratio. In this calculation the relative molecular backscattering distribution is determined from the remote atmospheric pressure and temperature measurements of the Jet Propulsion Laboratory (JPL) Microwave Temperature Profiler (MTP) along the flight track or from a meteorological analysis if the MTP data are not available. The aerosol scattering ratio is then just the total atmospheric scattering ratio minus one. Note that there is no attempt to correct for aerosol and cloud attenuation in these distributions.
Ten-day back trajectories for air parcels at 500 mbar along the flight track are also shown in Plate 2. These back trajectories were calculated using the Florida State University (FSU) kinematic back trajectory model. Air parcels along the flight track are selected at specific pressure altitudes, and these air parcels are followed back to their origins 10 days earlier (unless the trajectory terminates early at the surface). The consistency of the trajectories can be tested by using an ensemble of air parcels. This is discussed in detail by Fuelberg et at. [this issue]. The back trajectories shown in this paper are used to describe the general characteristics of the air mass' source region.

Air Masses Observed Over Pacific
A major air mass transition was observed during the survey flight from California to Hawaii on March 6, 1999 (flight 5). Plate 2 shows the 03 and aerosol distributions observed along the flight track as well as 10-day back trajectories arriving at 500 mbar along the flight track. The air off the coast of California had significant levels of 0 3 and enhanced aerosol loading across the entire troposphere. This situation gradually changed on the southwest flight leg such that the enhanced aerosol loading was no longer present and 03 levels near the surface were observed to be in the low 20 ppbv level. Cirrus clouds were seen with tops as high as 12 km which was just under the tropopause. A stratospheric intrusion was observed on the westbound leg with 03 levels over 60 ppbv extending down to 8 km at 20øN. This intrusion modified the 03 levels in the midtroposphere while 03 below 2 km was in the 20-40 ppbv range. The 10-day back trajectory analysis for air parcels arriving at 500 mbar (---5.6 km altitude) shows that the air closest to the California coast probably originated over Asia. The origin of the air along the track changed to coming from the southwest and then the west with less aerosols and 03 associated with it. This was a clear transition from a continen-tal source to more of a marine background condition with the additional influence of the stratospheric intrusion. A large contrast was found between the characteristics of air masses in the NH and SH across the Intertropical Convergence Zone (ITCZ). The meteorological conditions associated with PTB are described by Fuelberg et al. [this issue]. During PTB one segment of the ITCZ was located north of the equator between about 3 ø and 10øN. Plate 3 shows the 03, H20, and aerosol distributions across the ITCZ on the flight from Hawaii to Fiji on March 17-18, 1999 (flight 9). The transition from NH air to SH air can be seen near 2300 UT in the abrupt 03 decrease near the surface from 20-30 to less than 10 ppbv and in the midtroposphere from 40-60 to 20-40 ppbv. This decrease in 03 was anticorrelated with the change in H20 which generally increased everywhere below about 7 km. Convective outflow from upwind cloud activity can be readily seen in the H20 data between 2130 and 2230 UT at an altitude of 6.5 km. The depth of the marine boundary layer also increased into the SH air. There is some evidence of a slight enhancement in aerosols in the NH air (see aerosol layer at -4 km near 2130 UT). Cirrus clouds can be seen under the tropopause with the height of the tropopause increasing along the flight track from -13 km in the NH to over 15 km south of the equator. A series of trajectories arriving at three pressure levels (850, 500, 200 mbar) along this flight track are given in Plate 4. In the middle and lower troposphere north of about 10øN, the flow of the NH air was predominantly from Asia, while the flow to the south was entirely easterly near the surface and variable in the midtroposphere with light winds and long residence times over the Pacific. In the NH upper troposphere the flow was strongly from the west, while the SH flow was light and without organization. This is reflected in the vertical homogeneity of the SH air with low 03 and elevated H20 throughout the troposphere.  P'I-/PT9  N S  PT18  PT20 PT22PT26  PT28   I  I  I  I  I I  I  I  I  I  I 16.60 11.10 10.13 4.14 -8.00   (discussion in section 3.3). There is some indication that the 03 distribution in the lower troposphere is enhanced by the intrusion. This is not to ascribe all of the variation to this or another upwind intrusion; however, there is reason to attribute some of the 03 enhancement to a stratospheric source. We will address this later in this paper. Cirrus clouds are located on the equatorward side of the intrusion with cloud tops reaching the local tropopause altitude. The H20 distribution shows a moister troposphere on the north side of the intrusion and a generally drier atmosphere near the intrusion. The RH distribution shows the location of the intrusion better than the H20 distribution. The low RH values (<10%) associated with the intrusion are much smaller than the moderate (>40%) to high Composite average H:O and 03 profiles for PTB were calculated for the NH air north of 15øN, the SH tropical Pacific air between 10øN and 20øS, and the SH subtropical air south of 25øS with the results presented in Plate 14. The H20 is considerably higher in the SH tropical Pacific, particularly in the mid to upper troposphere. There is no statistically significant difference in our observed H:O profiles for the NH air and SH subtropical air above 2 km. Near the surface the SH subtropical air appears to be slightly drier than the NH air. The 03 profile for the SH tropical Pacific was lower than the other two regions at all altitudes. This profile starts at 12 ppbv near 1 km and does not exceed 28 ppbv until above l l km. The SH subtropical 03 profile is less than 10 ppbv greater than the SH tropical Pacific 03 profile up to about 8 km where the lower tropopause and stratospheric intrusions make a significant contribution at higher southerly latitudes. The NH 03 profile shows a more significant 03 enhancement throughout the mid to lower troposphere than either of the other profiles, and this difference was maintained compared to the SH tropical Pacific profile to above l l km.
Correlations between H:O and O3 were examined to better quantify differences in the three regions, and these results are shown in Plate 15. There was no statistical difference between the correlations at H:O levels greater than 1 g kg -• which is predominantly air below 5 km in the SH subtropics and below 7.5 km in the SH tropical Pacific. The correlation changed in the upper troposphere of the SH subtropics (>5 kin) as this air had more stratospheric influences. The H20 to 03 correlation was distinctly different for the NH air which was driven by the very different air mass origins. This was true for H:O greater than 0.1 g kg -• compared to SH tropical Pacific and greater than 1 g kg-• compared to SH subtropics. These results suggest a clear delineation between the SH and NH air masses with respect to H20 and 03, but the distinction between the SH subtropical air south of the SPCZ and the SH tropical Pacific air is more difficult to make on average possibly due to the limited sampling of air in the former category or the variability in the location or strength in the SPCZ.

3.3.
Air Mass Characterization 3.3.1. Air mass types and observations. A general categorization of air masses observed during PTB was done by using a variation of the principal component analysis on the 03 and aerosol distributions measured with the UV DIAL system and the PV values along the flight track derived from meteorological analyses from ECMWF data (details of this approach have been previously discussed by Browell et al. [1992Browell et al. [ , 1996aBrowell et al. [ , 1996b and Fennet al. [1999]). Since the background or reference air in the SH tropical Pacific during PTB was much different from that during PTA [Fenn et al., 1999 , air with high aerosol loading associated with boundary layer; clean Pacific (CP), ozone more than 20% below the reference profile and S < 0.2; convective outflow (CON), same as CP except cirrus clouds are in the vicinity; high-ozone plume (HPLU), ozone more than 20% above the reference profile and S > 0.2; high ozone (HO3), ozone more than 20% above the reference profile, S < 0.2, and amount of O3 attributable to stratosphere is <25%; high-ozone mixture (HO3M), same as HO3 except the amount of O3 attributable to stratosphere is 25-60%; stratospherically influenced (SINF), same as HO3 except that the amount of 03 attributable to stratosphere is >60%. Note that in the calculation of the amount of 03 attributed to the stratospheric component it is assumed that the O3 in the lower stratosphere has the following approximate relationship to PV:  bUnits of pptv/ppbv. cUnits of (fCi/scm)/ppbv. H20 , and RH were presented for this flight in Plate 8. The air mass classification method identified a large portion of the air above 2 km as having a strong stratospheric influence (SINF).

There are enhanced 03 regions where the stratospheric component explains a moderate amount of the enhanced 03 (HO3M), and a few small regions where there was very little stratospheric contribution (HO3). It is clear that the relatively low-resolution PV analyses cannot map all the features that the DIAL system observes. Thus there is some unavoidable misrepresentation of air masses between these groups in the vicinity of highly structured intrusions. This example also indicates air associated with deep convection with its associated low 03 and cirrus clouds (CON), near-surface air up to about 3 km (NS), and reference air (REF) predominantly on the equatorward side of the intrusion.
The results from each flight were grouped together based on the regions identified in Figure 1. The 5øN latitude line separating the NH regions from the SH low-latitude regions was chosen because it represented an approximate average location of the northern branch of the ITCZ. Also, several of the regions are the same as those used previously in the PTA air mass analysis [Fenn et al., 1999]

to permit comparisons between the field experiments. Plate 17 presents results for all of the PTB regions. The contrast in the air mass types identified in the NH (CPNH and EPNH) compared to the low-latitude cases (CPLL and EPLL) is readily apparent. There is a significant amount of enhanced 03 that is not attributed to stratospheric influence (HO3) in the NH.
In the EPNH the amount of enhanced 03 associated with enhanced aerosols increases significantly over the CPNH. There is some contribution from the stratosphere in the mid to low troposphere over the CPNH. At low latitudes over the central Pacific (CPLL) and eastern Pacific (EPLL), the situation is similar with the background or reference air dominating the distribution below 10 km. Additional contributions (up to ---20%) come from low 03 air associated with convective outflow (CON), enhanced 03 air resulting from photochemical production (HO3), and enhanced 03 air resulting from a mixture of photochemical 03 production and some stratospheric influence (HO3M). In the western Pacific low latitudes (WPLL), convective outflow of low 03 (CON) into the 3 to 13 km region is a major factor in determining the distribution of 03 in the mid to upper troposphere.  The midlatitude regions of the central Pacific (CPML) and the eastern Pacific (EPML) are very similar with the stratospherically influenced air (SINF) contributing to the 03 budget all the way down to the surface. While the SINF category was decreasing in frequency with decreasing altitude, the amount of HO3M was increasing significantly, which also indicates some stratospheric influence even though it is has been mixed with tropospheric air. The results from PTA (not shown) [Fennet al., 1999] indicate a dominant influence of enhanced 03 air masses (>60%) attributed to photochemical 0 3 production (HPLU and HO3) at SH midlatitudes in the western and central Pacific (WPML and CPML). These air masses also had a significant impact (>30%) in the midtroposphere even at low latitudes. Convective outflow of low 03 air from the surface to 7 to 15 km was frequently observed in the central Pacific low-latitude case during PTA, but during PTB it was not observed as often. This might be due to the smaller contrast in 03 distribution between the surface and the upper troposphere and an 03 discriminator that possibly made it more difficult to detect this type of air mass.
The average frequency of observation of all SH air mass types is presented in Plate 18 along with the average 0 3 profile for each air mass type. The reference air mass type occurred 44-53% of the time from 3 to 10 km, and the average 03 profile for it was very close to the discriminator 03 profile. This provided added support for this air mass type being the principal component in the analysis. The balance of the air mass types in this region were mostly distributed between convective outflow (CON), high 03 (HO3), and high 03 mixed (HO3M) with CON more prevalent in the upper altitudes and HO3 and HO3M more frequently observed in the lower free troposphere. As was expected, the near-surface air dominated the air mass types below 3 km, and in the upper troposphere the frequency of observations of stratospherically influenced air (SINF) increased with increasing altitude to the average tropopause level above 16 km where it was observed more than 86% of the time. There were few observations in the categories of aerosol plumes with background levels of 0 3 (BPLU) or of air with very low 0 3 air in the free troposphere that was not associated with convective outflow (CP). The average 03 profiles for the various air mass types show that at the lowest altitudes in the free troposphere (--•1-3 km) the SINF and HO3M categories have the highest average 0 3 levels associated with them, while above about 7 km, the SINF category has the highest observed 03 levels. The HO3 and HO3M categories have similar 03 profiles above 6 km, while below 6 km, HO3 is comparable to HPLU and NS at even lower altitudes. The CON 0 3 profile was constant at about 17 ppbv from about 5 to 13 km, and CP 0 3 profile was similar to CON with it decreasing to 5 ppbv at 1 km.
Average PV profiles for each air mass type are also shown in Plate 18. As expected, the SINF category had the highest average PV at all altitudes, and the HO3M category was intermediate between SINF and all the other air mass categories. REF, CON, and HO3 generally had an average of PVU -< 2 except for the REF profile above 9 km where it increased to PVU --•5 at 14 km. The explanation for this increase could be that the reference air in the upper troposphere could be a mixture of low 03 air from an upwind convective outflow with some stratospheric air that would bring the average 0 3 within the _+20% limits of the reference air 03 discriminator. For example, a mixing of seven parts of convective outflow air with 03 = 17 ppbv and PVU = 2 with one part of stratospheric air with 0 3 = 105 ppbv and PVU = 25 would produce air falling into the reference air category with 0 3 = 28 ppbv and PVU = 4.9. Using PV as a discriminator might provide insight into this mixing process. However, since PV does not have the spatial resolution of the 03 and aerosol measurements, this approach would be subject to even more uncertainties than the current method.
The PEM-Tropics B average O3/PVU ratio in high PVU (15-30) air designated as SINF was found to have a value of about 4.0. Even in the few cases encountered with 03 levels greater than 200 ppbv, the average O3/PVU ratio was about 3.2. These results are still in reasonable agreement with the stratospheric O3/PVU ratio of 4.2 estimated during PTA and used in the differentiation of PTB air masses involving enhanced 03 (HO3, HO3M, and SINF). The only effect of using the slightly higher ratio in the PTB analysis might be the slight overestimation of air mass observations in the HO3M and SINF categories as a result of slightly overestimating the 03 attributed to stratospheric air by less than about 15% at PVU values < 10.

Chemical air mass characterization.
A detailed in situ chemical characterization of each of the air mass types observed during PTB was determined from in situ measurements on the DC-8 [Raper et al., this issue] as the aircraft flew through the various air mass types that were remotely observed and categorized using the technique described above. This approach has been previously used by Browell et al. [1996aBrowell et al. [ , 1996b and Fennet al. [1999] in characterizing the chemical composition of air mass types observed during airborne field experiments. During PTB the average chemical characteristics of the various air mass types were determined for flights 5-21 with flight 22 omitted from the analysis since it was heavily influenced by air from North and South America. For the NS category, altitudes up to 1.8 km were included. Data segments were identified for each flight that contributed to the chemical characterization of various air mass types, and separate encounters with the same air mass type during a flight were counted as independent samples.

NH and SH air mass chemical characteristics:
Tables 1 and 2 present the average chemical compositions found for the major air mass types observed in NH and SH, respectively. This discrimination was necessary because of the different sources for the NH and SH air as discussed above.
The CON category is only listed for the SH air (Table 2)  A number of observations can be made concerning the chemical composition observed in the various air mass types in the NH (Table 1) and SH ( Table 2). The NH/SH ratios of several gases are given in Table 3 to aid in the comparison of the chemistry in the two hemispheres. All of the species listed in Table 3 Methyl iodide (CH3I) is relatively abundant in the marine boundary layer since the ocean is its source, and thus might serve as a tracer of marine convection Cohan et al., 1999]. As shown in Table 2, the average CH31 value was highest in the NS and CON air masses. The composition of CON was similar to the composition of NS, except for exhibiting much lower levels of H202, which is removed in cloud convective processes, and for having much higher ultrafine and unheated aerosol number densities, which may be associated with aerosol formation processes in cloud outflow regions.
Other  HNO 3, PAN, and NO•-) are associated with biomass burning, and they were generally lower during PTB than PTA. Ozone was significantly lower during PTB in all air mass types, and this is associated with a general reduction in photochemical production of 03 in the SH. Other species, such as CH 4 and CO2, were higher in all air mass types, and this is thought to be associated with the different seasonal background levels. Aerosol concentrations (UFA and FAU) were several times higher during PTB than during PTA. At least one aerosol precursor, SO2, was considerably more abundant in PTB than in PTA. The category ultrafine aerosol (UFA) includes aerosols with radii >8 nm, while the category fine aerosol unheated (FAU) includes aerosols with radii >18 nm. Since aerosols often form by nucleation and grow with time, the ratio FAU/ UFA is correlated with age of the aerosols, with a smaller ratio being associated with shorter ages, that is, a smaller fraction of larger particles. The characteristic times for growth from 8-18 nm to > 18 nm appear to be minutes to an hour or so [Hoppel et al., 1994]. If aerosols are still nucleating, the ratio will reflect that fact by being low. The ratios observed during PTB were only slightly higher than those observed during PTA, indicating that the short-term age of the air, in terms of aerosol formation, was not too different in the two missions.
Another indicator of air mass age is the ratio of C2H 2 to CO, with smaller values being associated with longer ages. C2H 2 has a lifetime of 0.041 years (15 days) [Ehhalt et al., 1998], which is much longer than aerosol growth rates. The PTB SH values were found to be significantly lower than the PTA values. The values (mean/median) for REF were 0.32/0.33 and 0.81/0.78 for PTB and PTA, respectively, 0.41/0.37 versus 1.09/1.13 for HO3M and 0.44 versus 1.14/1.15 for HO3. These values indicate that the air masses sampled during PTB were considerably older than those observed during PTA.

Summary and Conclusions
The PTB field experiment provided the first large-sc, ale characterization of air masses over the remote tropical Pacific Ocean during the austral late summer to early fall. The simultaneous UV DIAL-and LASE-derived measurements of 03, H20 , aerosols, and clouds across the troposphere along the flight track of the DC-8 aircraft allowed a more complete investigation of this area than ever before. PTB was conducted during a period of minimal biomass burning in the SH, and current results provide an important contrast to those of PTA which was conducted in the austral late winter to early spring when biomass burning was widespread in the SH. During PTB there was considerably less 03 than observed during the SH biomass burning period of PTA. The air mass characterization for PTA showed a large incidence in the high 03 (HO3) category (>60%) in the lower troposphere of central Pacific low latitudes compared to less than 24% in PTB. The enhanced 03 during PTA was a direct result of long-range transport of photochemically produced 03 from Africa and possibly as far away as Brazil.
The reduced 03 air observed during PTB was primarily confined to the SH tropical region from about 20øS to the ITCZ on the north and the SPCZ on the west. The moist easterly flow across the Pacific in the SH tropical region experienced progressive chemical 03 loss as it traveled to the western Pacific where 03 was found to be the lowest throughout the troposphere from less than 10 ppbv near the surface to less than 20 ppbv at 12 km. Ozone was generally negatively correlated with H20 across the Pacific with SH tropical air having generally less 03 and more H20 compared to NH air north of the ITCZ and SH subtropical air south of the SPCZ or •20øS. NH air had more cases of enhanced aerosols in plumes in the free troposphere than the SH air. The Asian/North American origins of the NH air and the Australian/African origins of the SH subtropical air were in contrast to the relatively slow easterly flow of SH tropical Pacific air that has not had continental influences from South America/Central America for many weeks. Conditions are optimum in the SH tropical Pacific air for photochemical loss of 03 (no continental pollution sources, high H20, and high solar insolation), and this is reflected in its average 03 profile with 12 ppbv at 1 km to about 28 ppbv from 7 to above 11 km.
Westerly flow in the middle and upper troposphere was stronger in the SH during PTA than PTB, and there was extensive biomass burning in Africa and other countries upwind of the South Pacific. Thus the resulting influence of bio-mXass burning sources on the remote tropical Pacific was much greater during PTA. A comparison of the average latitudinal and longitudinal 03 distributions for PEM-Tropics A and B provides clear evidence of the differences in flow conditions and chemistry between the two seasons. The latitudinal and longitudinal variations of the tropopause level can also be seen in the 03 distributions for the two missions. More stratospheric intrusions were observed during PTA than PTB; however, in both seasons, intrusions were found to extend to low latitudes (•20øS) in some cases. The observed intrusions in both field experiments contributed to determining the latitudinal structure of the tropopause and the 03 budget in the upper troposphere in the subtropics.
Air mass characteristics were determined for seven regions over the Pacific. Nine air mass types were identified for PTB based on their 03, aerosols, clouds, and PV characteristics. Data from each flight were categorized for each identified air mass, and these results were binned with altitude in the seven regions. NH air over the central and eastern Pacific had significantly different properties than SH air south of the ITCZ. The majority of the air masses in the NH were associated with O3-enhanced air in aerosol plumes or in low-aerosol air not associated with stratospheric intrusions. This is consistent with the 10-day back trajectories that originate over Asia. The SH low latitudes have mostly low 03 levels associated with reference or background air and even lower 03 associated with convective outflow and very low "clean Pacific" air. In midlatitudes the stratospheric influence is greater with more intrusions and the associated irreversible mixing of a portion of the stratospheric air into the troposphere. The altitude dependence of this observed process was also determined. The average composition for midtropospheric SH air was found to be dominated by the reference air (45-50%) which has a low 03 profile (<_28 ppbv) to 15 km. Convective outflow of very low 03 air from the surface, air with enhanced 03 due to photochemistry, and air with enhanced 03 due to a combination of photochemistry and stratospheric influences were found to be in comparable amounts (15-20% each) across this same region. Average 03 and PV profiles for each of the dominant air mass types were discussed with respect to their contribution to the observed chemical composition of SH air over the tropical Pacific, and an average O3/PVU ratio of about 4.0 was found in stratospheric air near the tropopause in good agreement with PTA observations. Major differences were observed in the chemical composition of the NH and SH air mass types with the NH air in the tropics being dominated by continental pollution transported from Asia. The air in the SH tropics had lower 03, CO, hydrocarbons, PAN, HNO3, and halocarbons compared to the NH air, and this air had lower C2H2/CO and C3Hs/C2H 6 ratios indicating older air. Composition of SH convective outflow air (CON) was similar to the near-surface (NS) air without the soluble species, but CON also had very elevated aerosol concentrations (UFA and FAU) which indicated considerable aerosol formation in the outflow regions.
Biomass burning influences on SH air during PTA dominated the chemical differences between PTA and PTB. PTB had greatly reduced 03, CO, hydrocarbons, PAN, HNO3, and nitrate due to low burning influences, but enhanced CH4, CO2, CH3I , and fine aerosols due to long-term trends and seasonal cycles. The SH air during PTB was considerably older in all air mass categories than the PTA air, and while the 03 was lower during PTB, the CO levels were even lower so that the O3/CO ratio was significantly lower for all the free tropospheric air mass types (REF, HO3, HO3M, and CON). The near surface (NS) category had nearly the same ratio as for PTA indicating that much of the impact of the PTA biomass burning was limited to the free troposphere. During PTB the average median O3/CO ratio of 0.35 in the near-surface (NS) air was found to be nearly the same as that found in the convective outflow air (CON) of 0.33, and this demonstrates the direct transport of air from near the surface to the mid to upper troposphere in cloud convective events.